Re: [Idr] draft-dong-idr-node-target-ext-comm-05.txt - WG Adoption and IPR call (9/27 to 10/11/2022)

[Robert Raszuk <[e-mail]>]

Hi Jeff,

You may call them "broken" or "bugs" - I would call it a tradeoff. And such
tradeoff for SAFI 128 and per VRF RD or consistent RT allocation actually
protocol wise works fine even if you filter on the sender post update
generation.

Customers are asking to speed BGP convergence and improve scale.
Throwing sticks in the update generation wheel does not help any of the
above. In fact it slows BGP down significantly.

So I believe this is what Gert and Randy were mainly bringing up. There is
not only risk of new bugs but features like this will have negative impact
on protocol and often irrespective if those new questionable features like
pushing config and policy by BGP in p2p manner would ever be used by ISPa,
ISPb or ISPc.

Regards,
R.


On Thu, Oct 20, 2022 at 11:00 PM Jeffrey Haas <[e-mail]> wrote:

> On Thu, Oct 20, 2022 at 10:53:11PM +0200, Robert Raszuk wrote:
> > Jeff,
> >
> > > P1 then is placed in 1.1.1.1's Adj-Rib-Out, but not 2.2.2.2's
> Adj-Rib-Out.
> > > Advertisement happens based on that state.
> >
> > That is one way it can be implemented.
> >
> > RTC has been implemented as filter post update generation in some
> operating
> > systems. In such cases your example does not apply.
>
> If such systems do not behave by withdrawing previously advertised state,
> they are broken.  Hopefully impacted customers raise the necessary bug
> reports.
>
> -- Jeff (already dealt with similar breakages across 3 stacks)
>